Dear all,
I experienced some problems with the efficiency of full reconstruction using the HIMSTER cluster of MAINZ:
1-The efficiency for the single particle reconstruction (electron in the present case) is about 10 % smaller that what I obtained in my previous simulations on my personal PC.
2-This efficiency obtained on HIMSTER changes by repeating the reconstruction step and using the same input from the event generator and the digitisation.

I used the scrut14 version of PANDAROOT compiled april13 external package.

After some debating with Anastasia, we realised the difference in the efficiency comes from pattern recognition step. This problem was difficult to debug, since it appears only in simulation at HIMster and we could not provide any test macros for external debugging. Therefore we went more deeply in the PndTrkTracking2 code by ourselves.
Among many other arrays which are defined, but not initialised, there is one which called statusflag. The values of this array is used during track search procedure. But since it is not initialised one can observe random or unstable behaviour. We would like kindly ask developers have a look at the code and fix it, please. As a temporary solution we initialised elements of the array by value 1. But we are not sure if this solution is correct one, because we don't know exactly the meaning of this array in the code.


Best regards,
Alaa & Anastasia.